Protecting Cloud Data: A Machine Learning Approach for Data Classification in Cloud Computing

Abstract

Cloud computing (CC) is a modern framework that enables users to store data on remote servers accessible via the internet. This model facilitates easy access and transfer of personal and critical data, leading to increased demand. Users can store various types of data, including financial transactions, documents, and multimedia content. Additionally, CC reduces reliance on local storage and lowers operational and maintenance costs. However, existing systems typically encrypt all data with the same key size, irrespective of its confidentiality level, resulting in higher processing costs and time. Moreover, these methods often classify data with low accuracy and fail to provide adequate confidentiality. This research introduces a cloud computing approach that employs automated data classification to assess data sensitivity. The proposed model categorizes data into three sensitivity levels: basic, confidential, and highly confidential. It utilizes Random Forest (RF), Naïve Bayes (NB), k-nearest neighbor (KNN), and Support Vector Machine (SVM) classifiers, incorporating automated feature extraction. The model achieved an accuracy of 92%, as demonstrated in simulation results. The findings indicate that RF, NB, and KNN outperform SVM. The research also offers valuable guidelines for cloud service providers (e.g., Dropbox and Google Drive) and researchers.